Hello Gert!

15 Oct 16, Gert Andersen wrote to Alexey Vissarionov:

 AV>> JFYI: when the installation program suggests to make separate
 AV>> partition for /boot, you'd better don't use the distribution built
 AV>> by idiots.

Just ignore comments that were done without knowledge about your system
configuration. There are good reasons for seperate /boot partitions. Some older
BIOS based machines can't boot too big harddisks without it.

 GA>>>>> but now will linux not mount it and says there is error on the
 GA>>>>> harddisk. It comes with that the superblok is missing o it

 GA> I still have all the files on my first node system

Backup is always good. The first action in broken HD faults is a disk copy with
the tool dd_rescue. dd_rescue ignores read errors and all readable data is
copied. If possible do another copy of that copy and use that copy for further
repair experiments. You don't know how long you could access the broken disk
and you don't know what damage your repair experiments do, so one "good" copy
should be always available.

If i had a working backup of the lost files i wouldn't put too much time into
rescue actions. But Alexeys steps are a good start.

 AV>>>> I'd like to have a look at the output of
 AV>>>> smartctl -A /dev/sdb \
 AV>>>> | sed -nre 's,^ +,,g;/^(5|187|188|197|198) /{s,  .* , ,g;p}'

 GA>>> I will see if I can do with this command

This gives you the S.M.A.R.T. status of the disk internal controller.
It filters relevant lines with sed. (add 195 to the list)

This is an example for a broken disk:

  5 Reallocated_Sector_Ct   0x0033   100   100   036    [...] 0
  7 Seek_Error_Rate         0x000f   082   060   030    [...] 186608327
195 Hardware_ECC_Recovered  0x001a   055   050   000    [...] 9
197 Current_Pending_Sector  0x0012   100   100   000    [...] 46
198 Offline_Uncorrectable   0x0010   100   100   000    [...] 46

 GA> I then after several testings and trying with function to repair and
 GA> then deleted all partions on it tried to see what happens by run
 GA> 'fdisk /dev/sdb'

So you already decided to risk all data. Get the real thing and throw the data
away, do a full badblock test in write mode over the full /dev/sdb (no number).
That may take some time but when finished you will know who much the disk is
broken.

 GA> and first make one large full big partion and it was not good when
 GA> gone to run e2fs.ext4 as the superblock not was written on the
 GA> harddisk

Then there must be badblocks on the area where the superblock should be
written.

 GA> , then I then tried to create to make 2 linux partions as 5mg and one
 GA> big by the reast of the hd and run e2fs.ext4 and this showed that the
 GA> superblock was made on the harddisk but it was bad done as the first
 GA> on the hd

It may be possible that the badblocks stay on the fist sectors of the disk and
do not grow. If so, the disk may be used further but should be considered
unsafe and stay under frequently observation. If the smartctl data grows,
discard the disk.

Tschuess

Kai

--- GoldED+/LNX 1.1.4.7